A UAV-Based System for Validating a Backward Lagrangian Stochastic Model in a Dairy Cattle Farm

Abstract

This study characterizes a compost-bedded pack barn of a dairy cattle farm in terms of CO2 emissions approximately 20 min after tilling under stable atmospheric conditions. Emission fluxes were calculated with the bLS model WindTrax, assessing modeled CO2 concentrations at two altitudes (5.0 m and 10.0 m ABGL) by comparing them with those measured by a UAV-based system at the same two altitudes. The UAV-based system was equipped with a low-cost self-engineered MSP (multi-sensor platform) containing an NDIR sensor for measuring concentrations and detecting environmental conditions, which were measured both by MSPs and commercial sensors. The input data were provided by the same sensors positioned on the ground (1.5 m ABGL), upwind and downwind with respect to the emission source. A sensitivity analysis of atmospheric stability in the bLS model yielded differences between median calculated emission fluxes for stable and unstable conditions from −0.020 to 0.034 g ∙ m−2 ∙ s−1. Mean percentage errors gave overestimates of 8–39% and 13–21% 5.0 m and 10.0 m ABGL. The RMSE also indicated overestimates ranging from 44 to 275 ppm. This is the first study to validate concentrations calculated by a bLS model at two altitudes by using a UAV-based system on a compost-bedded pack barn.